Lane 2: Approaches to AI-integrated assessment
AI can be integrated into assessment in different ways depending on the role AI plays in the learning task. In some assignments, AI may function primarily as a tool that supports students’ work, while in others it becomes the object of critical analysis or part of a collaborative workflow.
Choosing an appropriate approach
The most appropriate approach depends on the learning goals of the course and the role AI plays in the task. The approaches below describe different ways in which AI can function within an assignment, but they are not mutually exclusive. In practice, many assessments combine elements of multiple approaches. Across all approaches, the key principle remains the same: assessment focuses on students’ reasoning, judgement, and disciplinary understanding rather than on AI-generated output itself.
When selecting an approach, start by asking what students should ultimately learn to do.
| Learning goal | Matching approaches |
| Develop responsible use of AI tools within academic or professional workflows |
|
| Develop critical evaluation of AI systems and their limitations |
|
| Develop reasoning, argumentation, or decision‑making in interactive contexts |
|
Combining approaches
In many cases, assignments combine elements of several approaches. For example, students might use AI to generate an initial draft or set of ideas (approach 1), critically evaluate the quality and limitations of that output (approach 2), and document how their interaction with AI influenced the development of their work (approach 3). In other cases, students might interact with an AI-generated stakeholder or perspective (approach 4) and then analyse the strengths and weaknesses of the arguments presented.
By aligning the chosen approach with the intended learning outcomes, lecturers can design assessments that develop students’ ability to work critically and responsibly with AI while maintaining academic standards and disciplinary expectations.
Approach 1: AI as a tool
In this approach, AI tools support students’ work during tasks such as brainstorming, drafting, coding assistance, or structuring ideas. AI functions similarly to other research or writing tools: it may assist the process, but the student remains responsible for evaluating and refining the results.
Assessment therefore focuses on how students interact with and improve AI-supported work, rather than on the initial output produced by the tool. In this approach, AI supports the workflow, but the intellectual contribution remains clearly attributable to the student.
Assessment may examine:
- how students evaluated the quality or reliability of AI-generated material
- what revisions they made and why
- how disciplinary knowledge informed those revisions
- how the final work differs from the initial AI-generated output.
Example
Students use AI to generate an initial draft of a policy brief. They then revise the document using course literature and empirical evidence, explaining how the AI-generated material was evaluated, corrected, or expanded.
In this approach, AI supports the workflow, but the intellectual contribution remains clearly attributable to the student.
Approach 2: AI as an object of analysis
In this approach, AI-generated material becomes the object of critical examination. Students are asked to analyse, critique, or evaluate AI-produced responses using disciplinary concepts and theoretical frameworks.
Assignments of this type help students develop skills in evaluating the reliability, limitations, and biases of AI systems.
Examples include:
- identifying hallucinated references or factual inaccuracies
- comparing AI-generated summaries with original academic texts
- evaluating bias or omissions in AI responses
- critiquing AI-generated arguments using relevant theory.
Assessment focuses on:
- analytical depth
- use of disciplinary concepts or frameworks
- ability to identify limitations or inaccuracies in AI-generated material
- quality of reasoning in evaluating AI responses.
Example
Students compare an AI-generated explanation of a theoretical concept with the original scholarly source and analyse where the AI interpretation oversimplifies or misrepresents the argument.
Here, AI functions as a case study for critical analysis, rather than as a production tool.
Approach 3: AI as a collaborative partner in the workflow
In some assignments, students may work extensively with AI tools throughout the task. AI becomes part of an iterative research or problem-solving process, where students interact with the tool repeatedly and refine its outputs.
In this approach, assessment focuses heavily on the process of working with AI, rather than only on the final output.
Students may be asked to:
- document how prompts were developed and refined
- explain why particular outputs were accepted, rejected, or revised
- demonstrate understanding of the concepts involved in the task
- reflect on the limitations of AI-generated material
Assessment therefore emphasises:
- transparency about the workflow
- justification of decisions made during the process
- integration of disciplinary knowledge
- intellectual ownership of the final work.
Example
Students use AI to generate alternative methods to analysing a dataset. They then select, revise, and justify one method using course concepts and explain how AI suggestions were evaluated and modified.
In this approach, AI becomes part of the research process itself, but students remain responsible for interpreting results and making informed decisions.
Approach 4: AI as a simulation partner
In this approach, AI tools are used to simulate stakeholders, perspectives, or scenarios that students must analyse or respond to. AI becomes part of an interactive learning environment, allowing students to test arguments, explore alternative viewpoints, or practise professional decision-making.
The purpose of the assignment is not to evaluate the AI output itself, but to assess how students interpret, critique, and respond to the simulated interaction.
Examples include:
- engaging in a simulated policy negotiation with an AI-generated stakeholder
- questioning an AI system acting as a historical figure or theoretical perspective
- responding to AI-generated counterarguments in a debate exercise
- interacting with an AI system simulating a patient, client, or organisational decision-maker.
Assessment focuses on:
- the quality of the student’s reasoning and responses
- the use of disciplinary knowledge in analysing the simulated interaction
- the ability to identify weaknesses or limitations in AI-generated arguments
- the student’s reflection on how the interaction influenced their thinking.
Example
Students conduct a simulated negotiation with an AI system representing different policy stakeholders. They then analyse the interaction, identifying which arguments were persuasive, where the AI reasoning was flawed, and how their own position evolved during the exchange.
In this approach, AI functions as a dynamic stimulus for disciplinary reasoning, rather than as a source of content for students to reproduce.